Image generation

Drifting Models · 2026-02Natively few-step objectives

Learns a generator mapping directly, rather than a score or an instantaneous velocity to be integrated at sampling time. Training evolves the mapping's pushforward distribution along a drifting field until it matches the data dist…

RAE · 2025-10Representation-space generation

Replaces the VAE with a Representation Autoencoder: a frozen pretrained encoder (DINO, SigLIP, or MAE) plus a lightweight trained ViT decoder. The resulting latent is high-dimensional (width must be >= token dimension) and semanti…

PixNerd · 2025-07Single-stage pixel transformers

Replaces the pretrained VAE with a per-patch neural-field parameterisation, giving single-scale single-stage pixel diffusion with no tokenizer and no cascade, at ImageNet 256 FID 2.15 and 512 FID 2.84. Published four months before…

MeanFlow · 2025-05Natively few-step objectives

Replaces flow matching's instantaneous velocity target with an *average* velocity over a finite interval, giving a self-contained identity that supports one-step sampling trained from scratch with no distillation and no pretrained…

TarFlow · 2024-12Normalizing-flow revival

Introduces TarFlow, transformer-based autoregressive flow blocks with alternating autoregression direction, Gaussian noise augmentation during training, and a flow-specific guidance method. The architecture STARFlow later scales, …

OminiControl · 2024-11Adapter-based conditioningIn-context editing

Predates FLUX.1 Kontext by about seven months and generalizes the same core move beyond editing: reuse the DiT's own VAE encoder and transformer blocks for the condition image, concatenate condition and generation tokens into one …

Fluid · 2024-10Continuous-token autoregression

Scales the continuous-token autoregressive recipe to text-to-image at up to 10.5B parameters with random generation order, finding continuous tokens consistently better than discrete ones at every scale tested. The paper is carefu…

REPA · 2024-10Representation-aligned latents

Adds a simple auxiliary loss that aligns a diffusion transformer's intermediate hidden states to a frozen DINOv2's features during training. Speeds up SiT-XL training by 17.5x and sets a new ImageNet FID, with no change to the mod…

SiD2 · 2024-10Cascaded and multiscale pixel diffusion

Reaches FID 1.48 on ImageNet 512 in pixel space at 512 steps with guidance scale 1.0 over the interval (-3, 5), and 1.50 from a 16-step distilled variant. Competitive with the best latent baseline in a matched-setting table but na…

Autoguidance · 2024-06Guidance and sampling

Observes that classifier-free guidance conflates two things, improving image quality and narrowing diversity, and separates them by guiding with a deliberately weaker version of the same model rather than with an unconditional one…

MAR · 2024-06Continuous-token autoregression

Shows vector quantization is not required for autoregressive image generation: predict continuous per-token vectors and score them with a small conditional diffusion loss instead of a softmax over a discrete codebook, removing qua…

VAR · 2024-04Discrete-token autoregression

Redefines autoregressive image generation as coarse-to-fine next-*scale* prediction instead of raster-order next-token prediction; for the first time an AR image model beats diffusion transformers on ImageNet FID while also showin…

SD3 / MMDiT · 2024-03Flow matching and rectified flowVAE-latent diffusion

Takes rectified flow to production scale with a new dual-stream 'MM-DiT' architecture (separate weights for text/image tokens, joined by shared attention) plus a logit-normal timestep-sampling schedule; establishes the flow-matchi…

IP-Adapter · 2023-08Adapter-based conditioning

Moves adapter-based conditioning from spatial control to image-prompt conditioning: adds a second cross-attention branch with its own key/value projections just for image-prompt features, summed with the frozen model's existing te…

ControlNet · 2023-02Adapter-based conditioning

Freezes a pretrained UNet diffusion model entirely and adds a trainable copy of its encoder blocks, fed a spatial control signal (edges, depth, pose, segmentation) and connected back into the frozen backbone through zero-initializ…

DiT · 2022-12VAE-latent diffusion

Replaces the UNet backbone of latent diffusion with a plain Vision Transformer operating on latent patches, showing that transformer scaling laws (more FLOPs, lower FID) hold for diffusion just as they do for language modeling. Th…

Flow Matching · 2022-10Flow matching and rectified flow

Introduces a simulation-free objective for continuous normalizing flows: regress a vector field that generates any fixed conditional probability path between noise and data, with the straight-line (optimal transport) path as a sim…

Re-Imagen · 2022-09Agentic and search-augmented generation

Retrieves image-text pairs from an external knowledge base so the model can render rare and long-tail entities it has not memorised, and introduces a benchmark for exactly that failure. The paper frames this as an entity-memorisat…

Rectified Flow · 2022-09Flow matching and rectified flow

Independently derives the same straight-path ODE idea as flow matching, and adds 'reflow': iteratively re-coupling noise/data pairs along the learned ODE to make paths even straighter, enabling near-one-step sampling. The 'rectifi…

Classifier-Free Guidance · 2022-07Guidance and sampling

Trains one network on both conditional and unconditional objectives by randomly dropping the condition, then at sampling time extrapolates away from the unconditional prediction toward the conditional one. Removes the need for a s…

Imagen · 2022-05

A pre-2023, pixel-space (not latent) cascaded diffusion model, kept as a full entity despite D008 (see DECISIONS.md D018) because its conditioning-axis finding is load-bearing for 2023-2026 systems: a generic large language model …

LDM / Stable Diffusion · 2021-12VAE-latent diffusion

Moves diffusion training from pixel space into the latent space of a pretrained autoencoder, cutting compute by orders of magnitude while keeping quality; introduces the VAE + UNet + cross-attention recipe every 'latent diffusion'…

CDM · 2021-05Cascaded and multiscale pixel diffusion

Generates a low-resolution image and super-resolves it with a chain of pixel-space diffusion models, introducing conditioning augmentation to stop errors compounding along the chain. The origin of the position that pixel diffusion…

CLIP · 2021-02Contrastive language-image pretraining

Trains an image encoder and a text encoder jointly with a contrastive objective on 400M (image, text) pairs. Predates this atlas's 2023-2026 window and is not a diffusion paper, but its text tower became the first widely used prom…

LLaDA-Image · 2026-09VAE-latent diffusion

A 6B diffusion transformer trained from scratch alongside a frozen vision-language understanding module, released with weights, code and the full training recipe over 220 million samples — an unusually complete open release. The T…

GenFirst · 2026-08Semantic-plus-detail hybrids

Attacks the same latent-collapse problem as the representation-latent line but from the opposite direction: instead of freezing a semantic encoder, it trains generation-first under weak reconstruction pressure. Reports gFID 0.97 o…

Pixel T2I empirical study · 2026-08Single-stage pixel transformers

The first systematic industrial training recipe for pixel-space text-to-image, using a latent-to-pixel curriculum and reporting 3.2x to 4.8x inference speedup over latent baselines at comparable or better quality. Directly contrad…

Pixel-space survey · 2026-07Single-stage pixel transformers

A review covering architecture, noise scheduling and token efficiency across the pixel-space literature, arguing that a shared pixel, text and condition token space is a route to unified generation and understanding. Reports no ne…

Latent Diffusability · 2026-06Semantic-plus-detail hybrids

The empirical grounding the whole latent-design argument lacked: trains a large family of tokenizers spanning reconstruction-only, representation-only, and hybrid regularization, then measures which latent properties actually pred…

HiDream-O1-Image · 2026-05Editing inside a unified modelSingle-stage pixel transformers

Maps raw image pixels, text tokens and task-specific condition tokens into one shared token space processed by the same transformer blocks through joint self-attention, eliminating the VAE and the separate text encoder entirely. E…

HyperDiT · 2026-05Single-stage pixel transformers

Attacks the granularity dilemma with cross-attention to semantic anchors, scale-aware rotary embeddings across patch sizes, and register tokens taken from pretrained vision foundation models. ImageNet 256 FID 1.56. Concedes that q…

PAE · 2026-05Semantic-plus-detail hybrids

Asks the design question directly rather than picking a side: instead of inheriting a frozen foundation encoder (RAE) or training for reconstruction alone (VAE), it explicitly shapes a tokenizer's latent manifold using priors dist…

Qwen-Image 2.0 · 2026-05VAE-latent diffusion

The successor generation of Qwen-Image, extending fidelity and editing. Included because the Qwen-Image family is the open baseline most 2026 work measures against. Release status is unresolved: the technical report does not state…

Latent Forcing · 2026-02Semantic-plus-detail hybrids

Processes latents and pixels jointly under separate noise schedules, using the latent as a computational scratchpad rather than as the generative space. The explicit counter-position to the tokenizer-free line: not that latents ar…

Scale-RAE · 2026-01Representation-space generation

Tests whether the RAE recipe survives the jump from ImageNet to open-vocabulary text-to-image: scales a SigLIP-2-frozen RAE decoder on web, synthetic, and text-rendering data. Scale improves general fidelity but targeted data comp…

GenEval 2 · 2025-12Evaluation and benchmarks

Shows the field's most-cited compositional benchmark has drifted up to 17.7% away from human judgment as models improved, and proposes a corrected version. Directly relevant to any claim in this atlas that rests on a GenEval numbe…

DiP · 2025-11Single-stage pixel transformers

A DiT backbone for global structure plus a lightweight patch detailer head for local refinement, aimed squarely at the efficiency gap that justified latents in the first place. ImageNet 256 FID 1.79 at 631M parameters, and claims …

PixelDiT · 2025-11Single-stage pixel transformers

Splits the model in two: a patch-level DiT for global semantics and a pixel-level DiT that refines texture, end to end in raw pixels and extended to 1024px text-to-image. ImageNet 256 gFID 1.61 at 797M parameters. Its own limitati…

There is No VAE · 2025-10Single-stage pixel transformers

Pretrains an encoder self-supervised and then finetunes the whole thing end to end in pixels, reaching ImageNet 256 FID 1.58 at roughly 30 percent of DiT's training compute, and producing the first working pixel-space consistency …

Demystifying TM · 2025-10Transition matching

Independent theoretical analysis proving transition matching achieves strictly lower KL divergence than flow matching at finite step counts. Valuable to this atlas specifically because it is external validation rather than the ori…

DiffusionNFT · 2025-09Reinforcement learning and preference alignment

Optimizes the forward diffusion process directly instead of doing policy gradients through the reverse sampler, which makes it compatible with any black-box sampler and requires only clean images. An ICLR 2026 oral and a cleaner f…

HunyuanImage 3.0 · 2025-09Unified understanding and generation

An 80B mixture-of-experts model that unifies understanding and generation in one backbone, with autoregressive text generation and a diffusion head operating on continuous VAE latents for images, closer to Transfusion's hybrid tha…

Seedream 4.0 · 2025-09In-context editingVAE-latent diffusion

Folds text-to-image, editing, and multi-image composition into one system with native 4K output and dense text rendering. Representative of the industry-wide collapse of 'generation' and 'editing' into a single model rather than t…

NextStep-1 · 2025-08Continuous-token autoregression

Scales MAR's continuous-token idea to a 14B autoregressive model with a 157M flow-matching head, reaching state of the art among autoregressive text-to-image systems. An ICLR 2026 oral, and the strongest evidence that continuous-t…

Qwen-Image · 2025-08VAE-latent diffusion

A 20B MM-DiT whose distinguishing contribution is a data pipeline built for complex text rendering, an area every prior open system was visibly weak at. Uses a vision-language model as the text encoder rather than CLIP or T5, and …

REG · 2025-07Representation-aligned latents

Sharpens REPA's idea: rather than aligning hidden states to a frozen encoder as an auxiliary loss, it entangles a high-level semantic token directly into the denoising target alongside the low-level latents. A NeurIPS 2025 oral, a…

STARFlow · 2025-06Normalizing-flow revival

Revives normalizing flows as a serious high-resolution generator using a transformer autoregressive flow with a deep-shallow architecture, trained in a pretrained autoencoder's latent space. Notable for an explicit ablation findin…

Transition Matching · 2025-06Transition matching

Recasts generation as learning discrete-time Markov transition kernels over a continuous state, which unifies flow matching and continuous-token autoregression in one formulation instead of treating them as rival camps. Its fully-…

Flow-GRPO · 2025-05Reinforcement learning and preference alignment

Makes online reinforcement learning work for flow-matching models by converting the deterministic ODE into an SDE to recover the stochasticity policy-gradient methods need. Lifts SD3.5-Medium's GenEval score from 63% to 95%, and o…

PixelFlow · 2025-04Cascaded and multiscale pixel diffusionFlow matching and rectified flow

Cascaded flow matching directly in pixel space, ImageNet 256 FID 1.98 with no VAE. Carries the multiscale bet across to the flow objective, and lands between the two pixel lines.

REPA-E · 2025-04Representation-aligned latents

Shows naive end-to-end VAE+diffusion training collapses the VAE's latent space, but training through the REPA alignment loss instead (with a batch-norm bridge and a stop-gradient on the diffusion loss's path back to the VAE) makes…

IMM · 2025-03Natively few-step objectives

A single-stage few-step objective that matches distributions between marginals of a stochastic interpolant, giving distribution-level convergence guarantees that consistency models lack, with no pretraining or distillation. Reache…

VA-VAE / LightningDiT · 2025-01Representation-aligned latents

Names and measures the reconstruction-generation dilemma directly: a VAE tuned purely for reconstruction fidelity becomes a *harder* space for a diffusion model to learn in. Fixes it by adding a vision-foundation-model alignment l…

Edify Image · 2024-11Cascaded and multiscale pixel diffusion

An industrial-scale pixel-space system using Laplacian frequency bands, attenuating different bands at different rates. Evidence that a major vendor shipped pixel-space generation while the open literature treated latents as settl…

Adaptive Projected Guidance · 2024-10Guidance and sampling

Decomposes the guidance update into components parallel and orthogonal to the current prediction, and finds the parallel part causes oversaturation while the orthogonal part carries the prompt adherence. Suppressing the former kee…

Shortcut Models · 2024-10Natively few-step objectives

Conditions the network on the step size it is about to take, so a single model can jump accurately at any budget from many steps down to one, without a separate distillation stage or a second model.

GenArtist · 2024-07Agentic and search-augmented generation

A multimodal agent that plans a generation or edit, decomposes it into tool calls, and verifies the result before returning it. Bridges self-refinement loops to the current search-augmented wave.

CFG++ · 2024-06Guidance and sampling

Reframes guidance as an inverse problem and shows standard classifier-free guidance pushes samples off the data manifold at high scales, which is the mechanism behind the familiar oversaturation. The paper Rectified-CFG++ later ex…

Guidance interval · 2024-04Guidance and sampling

Shows guidance is actively harmful early in sampling and unnecessary late, so applying it only in a middle interval improves both sample and distribution quality. The direct conceptual ancestor of later work that treats when to gu…

HDiT · 2024-01Cascaded and multiscale pixel diffusion

A hierarchical transformer whose cost scales linearly rather than quadratically with pixels, making native megapixel pixel-space training feasible without a cascade or a latent. The architectural bridge between the cascaded era an…

SiT · 2024-01Flow matching and rectified flowVAE-latent diffusion

Reframes diffusion and flow matching as two points in one family of stochastic interpolants, isolating objective, coupling, and sampler as independent design choices on the exact DiT backbone; SiT-XL beats DiT-XL at matched size a…

GIVT · 2023-12Continuous-token autoregression

Removes the discrete codebook a different way from MAR: a linear projection in, and a Gaussian-mixture head out, so the transformer predicts continuous vectors without vector quantisation and without a diffusion loss. A sibling fo…

Idea2Img · 2023-10Agentic and search-augmented generation

Puts a multimodal model in a loop with an image generator so it drafts, critiques its own output and rewrites the prompt. The origin of the agentic half of this line, as distinct from the retrieval half.

T2I-Adapter · 2023-02Adapter-based conditioning

Published the same month as ControlNet with the same goal but a much smaller design — a compact external adapter network aligns internal T2I-model features with an external control signal while the base model stays frozen, giving …

simple diffusion · 2023-01Cascaded and multiscale pixel diffusion

Asks whether a single end-to-end pixel diffusion model can be made comparable to the two established alternatives, cascades and latent diffusion, using a shifted noise schedule and capacity concentrated at low resolution. It argue…

Parallel Decoding Distillation · 2026-07Natively few-step objectives

Trajectory-based few-step distillation that avoids the Jacobian-vector-product and finite-difference machinery MeanFlow-style objectives rely on, reporting state of the art at four to eight function evaluations.

Self-Sample Guidance · 2026-07Single-stage pixel transformersGuidance and sampling

Adds a lightweight self-guidance head on a frozen pixel-space backbone, improving JiT from 1.86 to 1.67 and PixelREPA from 1.81 to 1.59 for under one percent extra compute. An instance of the atlas's recurring finding that inferen…

Register Guidance · 2026-05Single-stage pixel transformers

Shows register tokens help pixel-space diffusion transformers considerably more than they help latent ones, and turns the observation into a guidance method. Evidence that pixel space is not simply latent space with more dimension…

STARFlow2 · 2026-05Normalizing-flow revivalUnified understanding and generation

Argues an autoregressive normalizing flow is structurally just an autoregressive transformer, then fuses a vision-language stream with a flow stream through residual connections so flow-generated visual tokens land in the same key…

PixelREPA · 2026-03Single-stage pixel transformersRepresentation-space generation

Finds that REPA, which reliably accelerates latent diffusion transformers, actively hurts JiT, and fixes it with a masked adapter. Valuable to the atlas as a negative result connecting two lines: representation alignment is not a …

Pixel MeanFlow · 2026-01Natively few-step objectivesSingle-stage pixel transformers

Carries MeanFlow into pixel space, removing the tokenizer entirely while keeping one-step generation. The clearest convergence point of the pixel-space and one-step lines: if it holds, neither a VAE nor a foundation encoder is nee…

SVG-T2I · 2025-12Representation-space generation

Scales SVG's frozen-DINOv3 recipe from ImageNet to text-to-image, reporting a 0.75 GenEval score and 85.78 DPG-Bench with full open-source release — the direct competitor to Scale-RAE, published about a month apart on essentially …

TM Design Space · 2025-12Transition matching

A systematic sweep of the transition-matching head across 56 text-to-image models at 1.7B scale, isolating which head and sampler combinations actually work.

JiT · 2025-11Single-stage pixel transformers

Argues pixel-space diffusion was never unstable because of pixels — it was unstable because of epsilon-prediction. Switching to x-prediction on large patches lets a plain ViT ('Just image Transformers') match tokenizer-based DiT/P…

Neon · 2025-10Reinforcement learning and preference alignment

Turns model collapse into a training signal: deliberately train on your own synthetic output, then extrapolate *away* from the degradation direction. An ICLR 2026 oral, and increasingly load-bearing as the open web fills with gene…

GLASS Flows · 2025-09Guidance and sampling

Simulates a flow model inside a flow model to get the sample quality of stochastic SDE sampling at the cost of deterministic ODE sampling, with no retraining. A pure inference-time win. The ICLR 2026 oral status reported in second…

Resurrect MAR · 2025-07Continuous-token autoregression

A systematic study of what actually limits masked autoregressive models, isolating the tokenizer and the bidirectional architecture as the binding constraints rather than the objective itself.

HiDream-I1 · 2025-05VAE-latent diffusion

A 17B sparse diffusion transformer released in fast, dev, and full variants — an open-weight system competitive with FLUX-class models, and one of the few using sparsity rather than dense scaling for efficiency.

GigaTok · 2025-04Discrete-token autoregression

Scales a visual tokenizer to 3B parameters and finds the reconstruction-versus-generation tension reappears at scale: better reconstruction does not monotonically improve downstream autoregressive generation unless the latent stay…

ICE-Bench · 2025-03Evaluation and benchmarks

Evaluates creation and editing jointly, matching the architectural shift toward single models that do both.

Lumina-Image 2.0 · 2025-03VAE-latent diffusion

An efficiency-oriented unified text-to-image framework using a Gemma-2 text encoder, reported to beat SD3 and Janus-Pro on several axes at lower cost. A common open efficiency baseline.

SANA-Sprint · 2025-03Natively few-step objectives

Combines continuous-time consistency distillation with an adversarial term to reach 1-4 step text-to-image generation, reported faster than FLUX-schnell at comparable quality. Distillation-based, so it depends on a strong multi-st…

DC-AE · 2024-10VAE-latent diffusion

Pushes autoencoder spatial compression up to 128x (vs. the standard 8x) without losing reconstruction quality, via residual autoencoding and a staged high-resolution adaptation schedule — up to 19x faster inference than SD-VAE-f8 …

Agentic Visual Generation · 2026-09Agentic and search-augmented generation

Surveys the emerging space where generative models are wrapped in an agentic control loop rather than called once, and proposes a taxonomy of control methods spanning image, video, 3D and interface generation.

PixSGR · 2026-09Single-stage pixel transformers

ImageNet 256 FID 1.51 and 512 FID 1.60, the lowest pixel-space numbers the atlas has found. One arXiv version old and unreviewed at time of writing, so it is recorded as a claim rather than as a settled result.

Any-OPD · 2026-08Natively few-step objectives

Distills a 12B model into a 2.5B one while treating the teacher as a black-box sampler, requiring no shared latent space between them — relevant to the atlas because it makes distillation independent of the representation choice.

ContextBias · 2026-08Evaluation and benchmarks

Tests whether stereotyped attributes persist when the prompt context changes, across 92 roles and more than 66,000 generated images from four current systems, and finds that they do.

MOSAIK · 2026-08Single-stage pixel transformers

Content-aware variable patch size, cutting 70 percent of FLOPs and 83 percent of tokens for a 1.0 point loss on DPG-Bench. Relevant because compute scaling with resolution is the standing objection to this whole line.

Observation Operators · 2026-08Single-stage pixel transformers

ImageNet 256 FID 1.52, among the strongest pixel-space results reported. Shares four authors with HyperDiT, so it is one group's continuing programme rather than independent confirmation.

Second-Order Drifting · 2026-08Natively few-step objectives

Diagnoses spectrally stiff convergence in Drifting Models, where fine detail resolves far more slowly than coarse structure, and lifts the dynamics into phase space with an auxiliary velocity term. Important caveat on scope: it is…

Chimera · 2026-07VAE-latent diffusion

A hybrid linear-attention and latent-attention diffusion transformer, with a scaling law fitted specifically for visual diffusion backbones rather than borrowed from language modelling.

SearchGen · 2026-07Agentic and search-augmented generation

Gives an image generator search access so it can render subjects its training data never covered, closing world-knowledge gaps rather than capability gaps. Releases a 20,000-example dataset and a benchmark.

CrossFlow · 2026-06Semantic-plus-detail hybridsNatively few-step objectives

A one-step flow from noisy latents to pixel-space images, FID 1.62 at a single function evaluation. A latent-to-pixel bridge rather than tokenizer-free generation, recorded here so the distinction stays visible.

Parallel Rollout Approximation · 2026-06Single-stage pixel transformers

Pixel-space autoregression rather than diffusion, reaching FID 1.94 at 511M parameters against a prior billion-scale pixel autoregressive FID of 3.60. Included because it tests this line's bet under a different generative objectiv…

Masked Bit Modeling · 2026-02Discrete-token autoregression

Moves autoregressive prediction down to the bit level rather than the codebook-entry level, sidestepping the vocabulary-size and codebook-collapse problems that constrain discrete token generation.

TM Distillation · 2026-01Transition matching

Applies transition matching to distil diffusion trajectories into few-step video generation. Recorded here as evidence the formulation travels beyond its originating group, with the caveat that the paper itself claims only the vid…

Rectified-CFG++ · 2025-10Guidance and sampling

Reworks classifier-free guidance specifically for rectified-flow models, where the original diffusion-era formulation is mismatched to straight-path transport and degrades samples at high guidance scales.

Tokenizer Post-Training · 2025-09Discrete-token autoregression

Argues tokenizers are trained on the wrong distribution: they see real images but must later decode *generated* latents. Post-training the tokenizer on the generative distribution closes that train/test gap and improves final samp…

Three-Body Scattering · 2026-07

A physics-inspired objective where each sample is attracted toward one real reference and repelled from an independently generated one. Reports 1.63 FID at one function evaluation on latent ImageNet 256. The authors position it ne…